Merge pull request #2786 from lonvia/export-centroid-for-tokenizer

[nominatim.git] / lib-php / TokenPartial.php

diff --git a/lib-php/TokenPartial.php b/lib-php/TokenPartial.php
index 477ef9c5a1547c149366844353b0f197edf1316b..3dc6f308a9e3c1f41e4e236163ccfc448973da48 100644 (file)
--- a/lib-php/TokenPartial.php
+++ b/lib-php/TokenPartial.php
@@ -1,4 +1,12 @@
 <?php
 <?php

+/**
+ * SPDX-License-Identifier: GPL-2.0-only
+ *
+ * This file is part of Nominatim. (https://nominatim.org)
+ *
+ * Copyright (C) 2022 by the Nominatim developer community.
+ * For a full list of authors see the git log.
+ */

 
 namespace Nominatim\Token;
 
 
 namespace Nominatim\Token;
 


@@ -8,16 +16,99 @@ namespace Nominatim\Token;
 class Partial
 {
     /// Database word id, if applicable.
 class Partial
 {
     /// Database word id, if applicable.

-    public $iId;
+    private $iId;

     /// Number of appearances in the database.
     /// Number of appearances in the database.

-    public $iSearchNameCount;
+    private $iSearchNameCount;
+    /// True, if the token consists exclusively of digits and spaces.
+    private $bNumberToken;

 
 

-    public function __construct($iId, $iSearchNameCount)
+    public function __construct($iId, $sToken, $iSearchNameCount)

     {
         $this->iId = $iId;
     {
         $this->iId = $iId;

+        $this->bNumberToken = (bool) preg_match('#^[0-9 ]+$#', $sToken);

         $this->iSearchNameCount = $iSearchNameCount;
     }
 
         $this->iSearchNameCount = $iSearchNameCount;
     }
 

+    public function getId()
+    {
+        return $this->iId;
+    }
+
+    /**
+     * Check if the token can be added to the given search.
+     * Derive new searches by adding this token to an existing search.
+     *
+     * @param object  $oSearch      Partial search description derived so far.
+     * @param object  $oPosition    Description of the token position within
+                                    the query.
+     *
+     * @return True if the token is compatible with the search configuration
+     *         given the position.
+     */
+    public function isExtendable($oSearch, $oPosition)
+    {
+        return !$oPosition->isPhrase('country');
+    }
+
+    /**
+     * Derive new searches by adding this token to an existing search.
+     *
+     * @param object  $oSearch      Partial search description derived so far.
+     * @param object  $oPosition    Description of the token position within
+                                    the query.
+     *
+     * @return SearchDescription[] List of derived search descriptions.
+     */
+    public function extendSearch($oSearch, $oPosition)
+    {
+        $aNewSearches = array();
+
+        // Partial token in Address.
+        if (($oPosition->isPhrase('') || !$oPosition->isFirstPhrase())
+            && $oSearch->hasName()
+        ) {
+            $iSearchCost = $this->bNumberToken ? 2 : 1;
+            if ($this->iSearchNameCount >= CONST_Max_Word_Frequency) {
+                $iSearchCost += 1;
+            }
+
+            $oNewSearch = $oSearch->clone($iSearchCost);
+            $oNewSearch->addAddressToken(
+                $this->iId,
+                $this->iSearchNameCount < CONST_Max_Word_Frequency
+            );
+
+            $aNewSearches[] = $oNewSearch;
+        }
+
+        // Partial token in Name.
+        if ((!$oSearch->hasPostcode() && !$oSearch->hasAddress())
+            && (!$oSearch->hasName(true)
+                || $oSearch->getNamePhrase() == $oPosition->getPhrase())
+        ) {
+            $iSearchCost = 1;
+            if (!$oSearch->hasName(true)) {
+                $iSearchCost += 1;
+            }
+            if ($this->bNumberToken) {
+                $iSearchCost += 1;
+            }
+
+            $oNewSearch = $oSearch->clone($iSearchCost);
+            $oNewSearch->addPartialNameToken(
+                $this->iId,
+                $this->iSearchNameCount < CONST_Max_Word_Frequency,
+                $this->iSearchNameCount > CONST_Search_NameOnlySearchFrequencyThreshold,
+                $oPosition->getPhrase()
+            );
+
+            $aNewSearches[] = $oNewSearch;
+        }
+
+        return $aNewSearches;
+    }
+
+

     public function debugInfo()
     {
         return array(
     public function debugInfo()
     {
         return array(


@@ -28,4 +119,9 @@ class Partial
                           )
                );
     }
                           )
                );
     }

+
+    public function debugCode()
+    {
+        return 'w';
+    }

 }
 }